New Normative Database of Inner Macular Layer Thickness Measured by Spectralis OCT Used as Reference Standard for Glaucoma Detection

Transcription

1 Article New Normative Database of Inner Macular Layer Thickness Measured by Spectralis OCT Used as Reference Standard for Glaucoma Detection María Nieves-Moreno 1,2, José M. Martínez-de-la-Casa 1,2, María P. Bambo 3, Laura Morales-Fernández 1,2, Karel Van Keer 4, Evelien Vandewalle 4, Ingeborg Stalmans 4, and Julián García-Feijoó 1,2 1 Ophthalmology Unit, Hospital Clínico San Carlos, Department of Ophthalmology, Universidad Complutense de Madrid, Madrid, Spain 2 Health Research Institute of the Hospital Clínico San Carlos, Madrid, Spain 3 Ophthalmology Unit, Hospital Miguel Servet, Zaragoza, Spain 4 Glaucoma Clinic UZ Leuven, Ophthalmology Research Group KU Leuven, Belgium Correspondence: José M. Martínezde-la-Casa, Hospital Clínico San Carlos, Calle del Profesor Martín Lagos s/n, Madrid, Spain. martinezcasa@ya.com Received: 17 October 2017 Accepted: 4 January 2018 Published: 28 February 2018 Keywords: ganglion cell layer; retinal nerve fiber layer; inner plexiform layer; Spectralis-OCT Citation: Nieves-Moreno M, Martínez-de-la-Casa JM, Bambo MP, Morales-Fernández L, Van Keer K, Vandewalle E, Stalmans I, García- Feijoó J. New normative database of inner macular layer thickness measured by Spectralis OCT used as reference standard for glaucoma detection. Trans Vis Sci Tech. 2018; 7(1):20, tvst Copyright 2018 The Authors Purpose: This study examines the capacity to detect glaucoma of inner macular layer thickness measured by spectral-domain optical coherence tomography (SD-OCT) using a new normative database as the reference standard. Methods: Participants (N ¼ 148) were recruited from Leuven (Belgium) and Zaragoza (Spain): 74 patients with early/moderate glaucoma and 74 age-matched healthy controls. One eye was randomly selected for a macular scan using the Spectralis SD- OCT. The variables measured with the instrument s segmentation software were: macular nerve fiber layer (mrnfl), ganglion cell layer (GCL), and inner plexiform layer (IPL) volume and thickness along with circumpapillary RNFL thickness (cprnfl). The new normative database of macular variables was used to define the cutoff of normality as the fifth percentile by age group. Sensitivity, specificity, and area under the receiver operating characteristic curve (AUROC) of each macular measurement and of cprnfl were used to distinguish between patients and controls. Results: Overall sensitivity and specificity to detect early-moderate glaucoma were 42.2% and 88.9% for mrnfl, 42.4% and 95.6% for GCL, 42.2% and 94.5% for IPL, and 53% and 94.6% for RNFL, respectively. The best macular variable to discriminate between the two groups of subjects was outer temporal GCL thickness as indicated by an AUROC of This variable performed similarly to mean cprnfl thickness (AUROC ¼ 0.845; P ¼ 0.29). Conclusions: Using our normative database as reference, the diagnostic power of inner macular layer thickness proved comparable to that of peripapillary RNFL thickness. Translational Relevance: Spectralis SD-OCT, cprnfl thickness, and individual macular inner layer thicknesses show comparable diagnostic capacity for glaucoma and RNFL, GCL, and IPL thickness may be useful as an alternative diagnostic test when the measure of cprnfl shows artifacts. Introduction Damage to the macula is common during the early stages of glaucoma. 1 However, this damage may be missed by both routine optical coherence tomography (OCT) and visual field (VF). Routine VF tests use test points spaced 68 apart. In the 24-2 VF test, for example, only four of the test points fall in the region of the macula, 1 which has the greatest density of ganglion cells. In patients with central VF loss, it has been shown that the thickness of the ganglion cellinner plexiform complex is a better variable to detect 1 TVST j 2018 j Vol. 7 j No. 1 j Article 20 This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

2 glaucoma than peripapillary retinal nerve fiber layer (RNFL) thickness. 2 Detecting macular thickness changes is important since have been recently reported that over 50% of eyes with mild to moderate glaucoma had defects within the central 638 of the visual field. 3 The latest software of the spectral-domain (SD) OCT instrument Spectralis allows for the automatic segmentation of the retinal layers but lacks a normative database for the thickness of each layer. Currently, a normative database of the thickness of each macular layer using Spectralis SD-OCT is not included in the device, and it would be useful to know if the inner macular layer thickness is within normal limits in glaucoma suspects or when peripapillary RNFL is not helpful due peripapillary abnormalities. In prior work, we generated a normative database for macular inner retinal layer thickness consisting of the volumes and thicknesses of the macular RNFL (mrnfl), ganglion cell layer (GCL), and inner plexiform layer (IPL). 4 For this study 300 healthy volunteers were recruited between the relatives and employees of Hospital Cĺinico San Carlos (Madrid, Spain) to develop the normative database. Every eye included in the study had a visual acuity equal or greater than 20/40, a sphere between 65 diopters (D) and a cylinder between 63 D, and no other ocular abnormalities. Macular OCT scans were performed on all patients with Spectralis OCT, the volume and the thickness of mrnfl, GCL, and IPL were analyzed with the Spectralis OCT segmentation software. The mean, 1st, 5th, and 95th percentile and the confidence interval of 95th and 99th were calculated for each variable to develop the normative database. The present study was designed to determine the diagnostic capacity of inner retinal layer thickness to detect glaucoma using this database generated as recommended by Realini et al. 5 to define the cutoff of normality. This diagnostic capacity was then compared with that of the thickness of the circumpapillary RNFL (cprnfl). Materials and Methods Subjects For this multicenter, observational, cross-sectional study, 148 eyes of patients with incipient to moderate glaucoma and control subjects were examined. All participants signed an informed consent form after listening to a detailed explanation of the tests and their purpose. The study protocol adhered to the tenets of the Declaration of Helsinki and received institutional review board approval from the Hospital Clínico San Carlos, Madrid (Spain), Hospital Miguel Servet, Zaragoza (Spain), and UZ Leuven, Leuven (Belgium). All subjects underwent a complete ophthalmic examination including biomicroscopy, measurement of intraocular pressure (IOP), refraction, and a fundus exam. Medical records were also examined. Inclusion criteria were: a visual acuity equal or greater than 20/40, a sphere between 65 D, and a cylinder between 63 D. Participants recently undergoing eye surgery (,1 year), with macular or retinal abnormalities and/or a medical history of neurological disease, diabetes, or any uncontrolled disease were excluded. Additional exclusion criteria for control subjects were a medical history of IOP.21 mm Hg or glaucoma. Only Caucasian individuals were selected for this study. For the patient group, 74 patients with primary open-angle glaucoma were recruited among the outpatients of the Glaucoma Clinic UZ Leuven and Hospital Miguel Servet, Zaragoza. Glaucoma diagnosis was performed by a glaucoma expert according to the criteria: glaucomatous optic neuropathy, IOP.21 mm Hg, and a visual field defect indicated by an Octopus perimetry (Haag-Streit, Mason, OH) test result of mean deviation (MD) from 3 to 9 db. Patients with secondary glaucoma were excluded. All patients underwent VF testing at least twice before their inclusion in the study. To define a VF defect as glaucomatous at least one of the following had to be detected in consecutive tests: having a cluster of three or more nonedge points showing a P, 0.05 and at least one point with a P, 0.01 in the pattern deviation probability plot, or pattern standard deviation of less than 5%. Reliability criteria were fixation losses of 20% or less, false-positive results of 15% or less, and false-negative results of 33% or less. For the control group, we recruited 74 healthy subjects visiting the Glaucoma Clinic UZ Leuven and Hospital Miguel Servet, Zaragoza for a routine eye test. These participants were selected following the recommendations of Realini et al. 4 In each participant, one eye was randomly selected for the final analysis provided both eyes met the inclusion criteria. An online statistical computing web programming was used to generate the randomization schedule. 2 TVST j 2018 j Vol. 7 j No. 1 j Article 20

3 Optical Coherence Tomography The Spectralis OCT (software version 6.0c; Heidelberg Engineering, Heidelberg, Germany) was used by an experienced operator to scan the macular and peripapillary regions on the same day as the rest of the tests without pupil dilation in a dark room. The automatic eye tracking technology of the Spectralis OCT maintains fixation on the retina, and only wellcentered images with a signal strength of.20 db were accepted. CpRNFL thickness measurements were made in the participants from Zaragoza (24 glaucoma patients and 24 controls) by conducting circular scans at a scanning angle of 128, assuming a standard corneal curvature of 7.7 mm; thus, projecting a 3.5-mm diameter circle on the retina. This full RNFL circle scan contained 768 A-scans along a peripapillary circle of The Spectralis software divides this circle into 6 regions: temporal, superotemporal, superonasal, nasal, inferonasal, and inferotemporal. Overall, RNFL thickness is provided as a number in the center. For the macular measurements in all participants, the scan was conducted on a cube with 49 raster lines separated by 120 lm, each containing 1046 pixels. Macular and inner retinal layer thicknesses were measured on 1-, 3-, and 6-mm rings as indicated in the macular map reported in the Early Treatment Diabetic Retinopathy Study (ETDRS). The 1-mm ring was defined as central thickness and the 3- and 6-mm rings as inner and outer rings each divided into the quadrants superior, nasal, inferior, and temporal. The numerical values for each of the nine subfields and macular volume were recorded and the device s segmentation software was then used to obtain individual retinal layer thickness measurements for the mrnfl, GCL, and IPL. Statistical Analysis Descriptive statistics are provided as mean 6 standard deviation. Mean macular RNFL, GCL, IPL, and peripapillary RNFL thicknesses were compared between control subjects and glaucoma patients using the Student s t-test for independent samples (applying Bonferroni correction). The diagnostic capacity of each variable to differentiate between healthy and glaucoma eyes was determined by calculating the sensitivity, specificity, positive predictive value, negative predictive value, positive and negative likelihood ratio, and diagnostic odds ratio. The cutoff used was the fifth percentile in Figure 1. patients. the normality database for each of four established age group (18 33, 34 50, 51 68, and years). We also calculated areas under receiver operating characteristic curves (AUROC) for two specific purposes: to assess the diagnostic power of each variable and compare the discrimination capacities of the macular and peripapillary variables. Finally, the differences between the macular variable AUROC and peripapillary variable AUROC were examined using the method of DeLong et al. 6 All statistical tests were performed using the software package SPSS 15 edition (SPSS, Inc., Chicago, IL). Significance was set at P, Results Histogram of perimetry mean defect for glaucoma The study participants (67 men, 81 women, all Caucasian) were recruited from the UZ Leuven Hospital (n ¼ 100) and Hospital Miguel Servet, Zaragoza (n ¼ 48). In these subjects, data for analysis were obtained for 78 right eyes and 70 left eyes. Glaucoma patients and control subjects were of similar mean age, years and years, respectively (P ¼ 0.7). The mean MD for the glaucoma patients was db (Fig. 1). Mean thickness data for mrnfl, GCL, IPL (n ¼ 148), and cprnfl (n ¼ 48) are provided in Table 1. When retinal layer thicknesses were compared between patients and controls, the major differences detected expressed as mean thickness reductions in glaucoma patients versus controls were: mrnfl outer inferior quadrant lm; GCL, inner temporal quadrant lm, outer temporal quadrant 9.84 lm, and inner inferior quadrant lm; and IPL, inner temporal quadrant 7.58 lm. For cprnfl, the greatest mean thickness reduction of 3 TVST j 2018 j Vol. 7 j No. 1 j Article 20

4 Table 1. Retinal Layer Thicknesses and Volumes Measured in Glaucoma Patients and Controls Parameter Control Glaucoma (MD. 9 db) P Macular RNFL Volume, mm ,0.001 Central, lm ,0.001 Inner nasal, lm Outer nasal, lm ,0.001 Inner superior, lm ,0.001 Outer superior, lm ,0.001 Inner temporal, lm Outer temporal, lm ,0.001 Inner inferior, lm ,0.001 Outer inferior, lm ,0.001 GCL Volume, mm ,0.001 Central, lm ,0.001 Inner nasal, lm ,0.001 Outer nasal, lm ,0.001 Inner superior, lm ,0.001 Outer superior, lm ,0.001 Inner temporal, lm ,0.001 Outer temporal, lm ,0.001 Inner inferior, lm ,0.001 Outer inferior, lm ,0.001 IPL Volume, mm ,0.001 Central, lm Inner nasal, lm ,0.001 Outer nasal, lm ,0.001 Inner superior, lm ,0.001 Outer superior, lm ,0.001 Inner temporal, lm ,0.001 Outer temporal, lm ,0.001 Inner inferior, lm ,0.001 Outer inferior, lm ,0.001 cprnfl Global, lm ,0.001 Temporal, lm Superior temporal, lm ,0.001 Inferior temporal, lm ,0.001 Nasal, lm ,0.001 Superior nasal, lm ,0.001 Inferior nasal, lm , lm was observed in the inferior temporal quadrant. The AUROC, sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios, and the diagnostic odds ratios calculated for each variable are shown in Table 2. To calculate all these indicators except AUROC, we used the fifth percentile obtained in the normative database. According to the cutoffs defined, the variables showing the best sensitivity (S) and specificity (Sp) to 4 TVST j 2018 j Vol. 7 j No. 1 j Article 20

5 Table 2. Glaucoma Diagnostic Capacity of Inner Macular Layers and Circumpapillary RNFL Using as Reference the New Normative Database Parameter AUROC Sensitivity % Specificity % VPP % VPN % LRþ LR OR Macular RNFL, n ¼ 148 Volume Central Inner nasal Outer nasal Inner superior Outer superior Inner temporal Outer temporal Inner inferior Outer inferior GCL, n ¼ 148 Volume Central Inner nasal Outer nasal Inner superior Outer superior Inner temporal Outer temporal Inner inferior Outer inferior IPL, n ¼ 148 Volume Central Inner nasal Outer nasal Inner superior Outer superior Inner temporal Outer temporal Inner inferior Outer inferior Peripapillary RNFL, n ¼ 48 Overall Temporal Superior temporal Inferior temporal Nasal Superior nasal Inferior nasal VPP, positive predictive value; VPN, negative predictive value; LRþ, positive likelihood ratio; LR, negative likelihood ratio; OR: diagnostic odds ratio. 5 TVST j 2018 j Vol. 7 j No. 1 j Article 20

6 Figure 2. ROC curve plots for mrnfl, GCL, IPL, and peripapillary RNFL. distinguish between glaucomatous and normal eyes were mrnfl volume (S ¼ 63.5%; Sp ¼ 93.2%), inner temporal GCL thickness (S ¼ 67.5%; Sp¼ 93.2%), inner inferior IPL thickness (S ¼ 54.1%; Sp ¼ 95.9%), and overall cprnfl thickness (S ¼ 70.8%; Sp ¼ 95.7%). According to the AUROC recorded for each layer examined, the outer inferior mrnfl zone (AUROC ¼ 0.880) showed the best diagnostic capacity, while worse results were obtained for the inner nasal and inner temporal mrnfl quadrants with AUROC close to 0.5. For the GCL, the best AUROC was obtained for the outer temporal quadrant (0.930) and all remaining zones except the central one showed AUROC over 0.7. For the IPL, the best AUROC was that recorded for the outer temporal quadrant (0.856) though again every subfield but the central one returned an AUROC greater than 0.7. The ROC curve plots for each layer are represented in Figure 2. No significant differences were found between the two best AUROC observed for the macular retinal inner layers and the AUROC recorded for mean overall cprnfl thickness. The results of this analysis are shown in Table 3. Discussion Recent efforts have been made to develop new imaging techniques that will help detect glaucoma in its early stages. One of these techniques, SD-OCT, has served to identify new structural parameters to measure RNFL thickness in the peripapillary area, as well as retinal thickness and retinal inner layer thicknesses in the region of the macula. 7 In our study, significantly reduced thicknesses were observed in glaucoma patients compared with healthy controls in most inner retinal layer zones of the macular area (mrnfl, GCL, and IPL). It is important to study this differences to understand how glaucoma affects the macula and it may be helpful diagnosing glaucoma in an earlier stage. In the inner temporal zone our mrnfl thickness measurements indicated a thickness increase of TVST j 2018 j Vol. 7 j No. 1 j Article 20

7 Table 3. Nieves-Moreno et al. Comparing Areas Under the ROC Curves Among the Best Discriminatory Parameters Comparison Between Difference Between AUROC P Overall cprnfl thickness Outer superior mrnfl thickness mrnfl volume Outer temporal GLC thickness GCL volume Inner temporal IPL thickness Outer temporal IPL thickness lm for glaucoma patients versus controls. This increase, despite being statistically significant (P ¼ 0.03), is not clinically relevant. The most pronounced mrnfl thickness reduction detected was that noted for the outer inferior quadrant (mean reduction lm). Other authors such as Hood et al. 1 have also reported greatest RNFL thickness reductions in glaucomatous eyes in the arcuate regions, with greatest thinning produced in the outer inferior zone. Consistent with the present findings, in the study by Hood et al. 1 greater thinning of the GCL-IPL complex was detected in the temporal quadrants of the macula. Also in line with our measurements, Tan et al. 8 noted greatest ganglion cell complex thickness reductions in preperimetric glaucoma patients versus controls in the temporal quadrants. Similarly, a recent study has shown GCL and IPL thinning when these layers were measured individually by Spectralis OCT in patients with glaucoma compared with healthy individuals. 9 When comparing the glaucoma diagnostic capacity of the retinal inner layers, the mrnfl measurement that showed the best AUROC was the outer inferior quadrant. Inner nasal and inner temporal mrnfl thicknesses offered the worse diagnostic power indicated by AUROC close to 0.5. Consequently, we can state that this inner region is barely able to discriminate between healthy subjects and glaucoma patients. In a similar study comparing glaucoma suspect patients and controls, the best AUROC for mrnfl thickness were obtained for the quadrants inner temporal (AUROC ¼ 0.742) and outer temporal (AUROC ¼ 0.660). 9 These findings are in agreement with the present results and are also in line with histologic observations of the death of the ganglion cell body when the axon degenerates. 10 Among our GCL thickness measurements, those that showed the best AUROC were the measurements taken in the outer temporal and remaining subfields except the central zone that showed AUROC over 0.7. For IPL thickness, the highest AUROC was obtained for the outer temporal and, again, areas under the curve for every zone but the central were greater than 0.7. These results are similar to those obtained in the study by Chien et al., 9 in which best AUROC were obtained in the outer temporal quadrant, both for GCL and IPL thickness (0.942 and 0.890, respectively). The best AUROC obtained by Chien et al. 8 is slightly higher than the one obtained in our study. This small difference could be attributed to the glaucoma patients in the Chien et al. 8 study showing a greater mean visual field defect than our glaucoma patients (MD ¼ db vs db). This difference is important because it could indicate that the diagnostic power of the test improves with glaucoma severity, as occurs with most diagnostic tests. The GCL and IPL thinning found here in patients with glaucoma is consistent with the thickness reductions recently reported by Lee et al., 11 who observed greater AUROC for Spectralis-OCT IPL plus GCL thickness in the outer temporal and outer inferior quadrants. The sensitivity and specificity of the cutoffs defined by the new database of retinal layer thicknesses by macular zone are similar to those reported by others. In the Cochrane library review, the sensitivities and specificities of several GCC and GCL-IPL parameters reported in 35 studies measured with the instruments Topcon 3D OCT, RTVue OCT, and Cirrus HD OCT were in the same range as observed for cprnfl thickness measurements. 12 In our study, except for measurements in the IPL nasal quadrants, inner layers thickness in the outer zones showed a better diagnostic capacity than thickness measurements made in the inner quadrants of the macula. This is attributable to a greater susceptibility to glaucomatous damage of the outer arcuate fibers followed by the superior arcuate fibers, which encompass the outer areas of the ETDRS subfields. In contrast, the papillomacular bundle, which contains fibers from the central ETDRS zones, remains unaffected until advanced stages of glauco- 7 TVST j 2018 j Vol. 7 j No. 1 j Article 20

8 ma. This vulnerability of the inferior arcuate fibers can be explained by their introduction in the optic nerve through the inferior quadrant of the optic disc, which is the first affected structure in this progressive optic neuropathy. In contrast, the superior arcuate fibers insert in the optic nerve through the temporal zone. Thus, this region is unaffected until more advanced stages of glaucoma. 1 No significant differences were found here between the best AUROC obtained for the macular parameters and the AUROC obtained for overall cprnfl thickness. This finding is in agreement with the findings of the Cochrane library review, 12 in which no differences emerged among the diagnostic capacities of retinal inner layer thickness, optic nerve head parameters and cprnfl thickness. However, it should be stressed that no Spectralis SD-OCT measurements were included in the Cochrane review. Martínez de la Casa et al. 13 reported a significant difference when comparing AUROC for macular RNFL thickness measured in the outer temporal quadrant and the cprnfl variable showing the best diagnostic capacity, temporal cprnfl thickness (difference between both AUROC ¼ 0.147, P ¼ 0.042). In contrast, in a recent study published by Pazos et al., 14 though macular variables showed a high diagnostic power, the diagnostic capacity of both macular and peripapillary parameters was comparable. This differences in the results may be due to the difference on the glaucoma stage of the patients, because the patients included by Martinez de la Casa et al. were glaucoma suspects, and ours and Pazos et al. patients had established glaucoma. In a recent review, 15 the diagnostic capacities of macular and peripapillary measurements made by RTVue, Topcon 3D, and Cirrus OCT reported in 34 studies (n ¼ 2164 participants) were compared. The conclusion of this review was that using the methods available today, cprnfl thickness measurements are better than macular thickness measurements to detect glaucoma though the differences detected using these OCT techniques were small. Our study has several limitations. As it is a casecontrol study, it could overestimate the diagnostic capacity of the tests used. Also, the age group of the subjects was taken into account when establishing the cutoff point for glaucoma, whereas sex was not considered. Several studies have shown thickness differences between males and females in the majority of quadrants possibly leading to the incorrect classification of subjects with normal, borderline, or abnormal retinal layer thickness measurements. Another limitation of our study was that only Caucasian patients were included, such that results might not be the same for patients of other ethnicities Only patients with moderate refractive errors were include in this study, so ours results may not be applicable to patients with high refractive errors. This measurements where done with three different Spectralis OCT devices but small changes may be found when using another device. Further, as it was a cross-sectional study, neither patients nor controls were followed after the Spectralis OCT measurements. Therefore, some control subjects may have developed glaucoma in the months following their inclusion in the study. Finally, the sample size used to compare the diagnostic capacities of retinal inner layer thicknesses and peripapillary RNFL thickness was small (n ¼ 48), so discrete differences could have been undetected. Currently, Spectralis SD-OCT does not provide any data about the normality or abnormality of the macular layer thickness, and it only shows a thickness distribution map of every layer. To the authors knowledge, this is the first study to analyze the diagnostic capacity of the inner macular layer thickness using a normative database, and including the normality values in the device would made the results interpretation easier. In summary, as measured by Spectralis SD-OCT, cprnfl thickness and individual macular inner layer thicknesses show comparable diagnostic capacity for glaucoma. Notwithstanding, it has been suggested that macular OCT might perform better in patients with central VF defects than in patients with peripheral defects. 2 There is, therefore, a need for further studies on this topic in which the pattern of the VF defect is also considered. Acknowledgments Disclosure: M. Nieves-Moreno, None; J.M. Martínez-de-la-Casa, None; M.P. Bambo, None; L. Morales-Fernández, None; K. Van Keer, None; E. Vandewalle, None; I. Stalmans, None; J. García- Feijoó, None References 1. Hood DC, Raza AS, de Moraes CG V, et al. Glaucomatous damage of the macula. Prog Retin Eye Res. 2013;32: TVST j 2018 j Vol. 7 j No. 1 j Article 20

9 2. Shin HY, Park HYL, Jung KI, et al. Glaucoma diagnostic ability of ganglion cell-inner plexiform layer thickness differs according to the location of visual field loss. Ophthalmology. 2014;121: Schiefer U, Papageorgiou E, Sample PA, et al. Spatial pattern of glaucomatous visual field loss obtained with regionally condensed stimulus arrangements. Invest Ophthalmol Vis Sci. 2010; 51: Nieves-Moreno M, Martínez-de-la-Casa JM, Cifuentes-Canorea P, et al. Normative database for separate inner retinal layers thickness using spectral domain optical coherence tomography in Caucasian population. PLoS One. 2017;12: e Realini T, Zangwill LM, Flanagan JG, et al. Normative databases for imaging instrumentation. J Glaucoma. 2015;24: DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44: Mwanza J-C, Budenz DL, Godfrey DG, et al. Diagnostic performance of optical coherence tomography ganglion cell inner plexiform layer thickness measurements in early glaucoma. Ophthalmology. 2014;121: Tan O, Chopra V, Lu ATH, et al. Detection of macular ganglion cell loss in glaucoma by Fourier-domain optical coherence tomography. Ophthalmology. 2009;116: e2. 9. Chien JL, Ghassibi MP, Patthanathamrongkasem T, et al. Glaucoma diagnostic capability of global and regional measurements of isolated ganglion cell layer and inner plexiform layer. J Glaucoma. 2017;26: Quigley HA. Neuronal death in glaucoma. Prog Retin Eye Res. 1999;18: Lee KM, Lee EJ, Kim TW, Kim H. Comparison of the abilities of SD-OCT and SS-OCT in evaluating the thickness of the macular inner retinal layer for glaucoma diagnosis. PLoS One. 2016;11:e Michelessi M, Lucenteforte E, Oddone F, et al. Optic nerve head and fibre layer imaging for diagnosing glaucoma. Cochrane Database Syst Rev. 2015;11: CD Martinez-de-la-Casa JM, Cifuentes-Canorea P, Berrozpe C, et al. Diagnostic ability of macular nerve fiber layer thickness using new segmentation software in glaucoma suspects. Invest Ophthalmol Vis Sci. 2014;55: Pazos M, Dyrda AA, Biarnés M, et al. Diagnostic accuracy of Spectralis SD OCT automated macular layers segmentation to discriminate normal from early glaucomatous eyes. Ophthalmology. 2017;124: Oddone F, Lucenteforte E, Michelessi M, et al. Macular versus retinal nerve fiber layer parameters for diagnosing manifest glaucoma: a systematic review of diagnostic accuracy studies. Ophthalmology. 2016;123: Appukuttan B, Giridhar A, Gopalakrishnan M, Sivaprasad S. Normative spectral domain optical coherence tomography data on macular and retinal nerve fiber layer thickness in Indians. Indian J Ophthalmol. 2014;62: Massin P, Erginay A, Haouchine B, et al. Retinal thickness in healthy and diabetic subjects measured using optical coherence tomography mapping software. Eur J Ophthalmol. 2002;12: Won JY, Kim SE, Park Y-H. Effect of age and sex on retinal layer thickness and volume in normal eyes. Medicine (Baltimore). 2016;95: e Kelty PJ, Payne JF, Trivedi RH, et al. Macular thickness assessment in healthy eyes based on ethnicity using stratus OCT optical coherence tomography. Invest Ophthalmol Vis Sci. 2008;49: Grover S, Murthy RK, Brar VS, Chalam KV. Normative data for macular thickness by highdefinition spectral-domain optical coherence tomography (Spectralis). Am J Ophthalmol. 2009; 148: Chauhan DS, Marshall J. The interpretation of optical coherence tomography images of the retina. Invest Ophthalmol Vis Sci. 1999;40: TVST j 2018 j Vol. 7 j No. 1 j Article 20